PE-to-PE latency test + supporting fixes

Adds tests/test_pe_to_pe_latency.py: a sweep that measures PE-to-PE
transfer latency for five hop types (intra-cube horizontal/vertical,
inter-cube horizontal/vertical, inter-SIP) across data sizes 128 B to
10 KB, on both the IPCQ (tl.send/tl.recv) and raw-DMA (tl.load+tl.store)
paths. Emits per-hop PNG plots, an overview PNG, and a CSV summary into
tests/pe2pe_latency_plots/. Latency is reported as max(pe_exec_ns) across
participating PEs, read from engine.get_completion(), so the measurement
captures the SRC/DST PE's kernel body time rather than the full launch+
response-aggregation envelope.

Two simulator fixes were needed to make this measurement meaningful:

- PeMMU now stores a list of (start, end, pa) sub-regions per page
  rather than a single PA. DPPolicy layouts with shards smaller than
  page_size (e.g. 128 B payloads with 4 KB pages) used to silently
  overwrite each other through last-write-wins, causing DMAs intended
  for cube0 to physically route to cube3 - inflating latency by ~170 ns
  per DMA at small sizes. STOPGAP: real MMUs don't support sub-page
  regions; long-term fix is either smaller MMU page size or DPPolicy
  validation that refuses sub-page shards.

- M_CPU's per-PE metrics aggregation (pe_exec_ns, dma_ns, compute_ns)
  now max-merges against the existing value in result_data rather than
  overwriting. Multi-cube workloads share one result_data dict via
  IO_CPU fanout; the previous overwrite caused whichever cube's M_CPU
  finished last to clobber others' values, so multi-cube pe_exec_ns was
  racy and frequently 0. Same fix applied in legacy/builtin/m_cpu.py.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

src/kernbench/components/builtin/m_cpu.py

@@ -204,16 +204,21 @@ class MCpuComponent(ComponentBase):
         yield all_done
         del self._parent_txns[request.request_id]
 
-        # Aggregate PE-internal metrics (max across PEs)
+        # Aggregate PE-internal metrics (max across PEs and across cubes).
+        # Multiple M_CPUs share the same result_data dict via IO_CPU fanout;
+        # merge against the existing value so cubes don't clobber each other.
         pe_exec_values = [st.result_data.get("pe_exec_ns", 0.0) for st in sub_txns]
         if pe_exec_values:
-            txn.result_data["pe_exec_ns"] = max(pe_exec_values)
+            cur = txn.result_data.get("pe_exec_ns", 0.0) or 0.0
+            txn.result_data["pe_exec_ns"] = max(cur, max(pe_exec_values))
         dma_values = [st.result_data.get("dma_ns", 0.0) for st in sub_txns]
         if dma_values:
-            txn.result_data["dma_ns"] = max(dma_values)
+            cur = txn.result_data.get("dma_ns", 0.0) or 0.0
+            txn.result_data["dma_ns"] = max(cur, max(dma_values))
         compute_values = [st.result_data.get("compute_ns", 0.0) for st in sub_txns]
         if compute_values:
-            txn.result_data["compute_ns"] = max(compute_values)
+            cur = txn.result_data.get("compute_ns", 0.0) or 0.0
+            txn.result_data["compute_ns"] = max(cur, max(compute_values))
 
         # Send aggregate response on reverse command path back to IO_CPU
         reverse_path = list(reversed(txn.path))

src/kernbench/components/legacy/builtin/m_cpu.py

@@ -204,16 +204,21 @@ class MCpuComponent(ComponentBase):
         yield all_done
         del self._parent_txns[request.request_id]
 
-        # Aggregate PE-internal metrics (max across PEs)
+        # Aggregate PE-internal metrics (max across PEs and across cubes).
+        # Multiple M_CPUs share the same result_data dict via IO_CPU fanout;
+        # merge against the existing value so cubes don't clobber each other.
         pe_exec_values = [st.result_data.get("pe_exec_ns", 0.0) for st in sub_txns]
         if pe_exec_values:
-            txn.result_data["pe_exec_ns"] = max(pe_exec_values)
+            cur = txn.result_data.get("pe_exec_ns", 0.0) or 0.0
+            txn.result_data["pe_exec_ns"] = max(cur, max(pe_exec_values))
         dma_values = [st.result_data.get("dma_ns", 0.0) for st in sub_txns]
         if dma_values:
-            txn.result_data["dma_ns"] = max(dma_values)
+            cur = txn.result_data.get("dma_ns", 0.0) or 0.0
+            txn.result_data["dma_ns"] = max(cur, max(dma_values))
         compute_values = [st.result_data.get("compute_ns", 0.0) for st in sub_txns]
         if compute_values:
-            txn.result_data["compute_ns"] = max(compute_values)
+            cur = txn.result_data.get("compute_ns", 0.0) or 0.0
+            txn.result_data["compute_ns"] = max(cur, max(compute_values))
 
         # Send aggregate response on reverse command path back to IO_CPU
         reverse_path = list(reversed(txn.path))